As of today, we’ve added a simulation offering for both IonQ Aria and IonQ Harmony to The IonQ Quantum Cloud. This expands our already-extensive developer tooling, allowing algorithm developers to better understand and design for the specific characteristics of IonQ hardware when prototyping quantum solutions.
By adding a few new optional parameters to your simulation requests on our cloud simulator, you can run a hardware simulation, a quantum circuit simulation that behaves more like if it was run on real hardware. This is accomplished via a calibrated noise model—a simplified fingerprint of the characterized noise channels—that allows the simulator to approximate the results of running on a specific hardware system.
Initial hardware noise models are now available for IonQ Aria and Harmony, with a noise model for IonQ Forte coming soon, and improvements to noise models rolling out as we continue to characterize and model noise in systems. They are free to use for anyone using the IonQ API directly, including those using it through our Google Cloud Marketplace integration. For more details on how to run a hardware noise model simulation, read our getting started guide.
The Value of Noise
“Noisy” simulation may seem like an undesirable quality, but it’s actually highly useful to algorithm developers: by default, a quantum simulator provides an “ideal” or noiseless simulation, where the simulator behaves as a hypothetical “perfect” quantum computer, with no sources of error. While this type of simulation is invaluable in understanding and characterizing the precise expected behavior of an algorithm, it lacks one thing: it doesn’t tell us much about how a real near-term system might run the circuit.
Because all near-term systems — even the world’s most powerful — have a variety of non-uniform sources of noise, algorithms run on hardware will not be ideal and when an intrepid algorithm developer moves from simulation to hardware, the outcome changes in potentially-unexpected ways. Certain noise characteristics may interact with an algorithm in specific ways, and without designing with these in mind, developers can’t get the most out of near-term hardware.
System noise model simulations unlock a new ability for quantum researchers and algorithm developers to better understand how circuits will run on-system before they actually do so. While these simplified noise fingerprints are not a complete replacement for running certain algorithms on hardware — we wouldn’t, for example, recommend using noise model simulation to train an error mitigation scheme, or as a precise prediction of what a system can or can't do — they are still very valuable in improving developer workflows and understanding by being able to understand the impact of noise on algorithms. To learn more about how to get started with noise model simulation, check out our new guide: Get Started With Hardware Noise Model Simulation.